• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site
Menu
  • HSE University
  • Publications of HSE
  • Articles
  • Интеграция предиктивных моделей в состав платформ интернета вещей для реализации сценариев управления энергопотреблением в режиме на сутки вперед

Article

Интеграция предиктивных моделей в состав платформ интернета вещей для реализации сценариев управления энергопотреблением в режиме на сутки вперед

Датчики и системы. 2020. № 11. С. 19-29.
Кычкин А. В., Горшков О. В.

The task of developing the functionality of typical Internet of Things (IoT) platforms to the level of using custom predictive models in energy management of buildings, structures and industrial facilities in the day-ahead mode is considered. Predictive control scenarios of both single loads (consumers) and their aggregated groups can be used to reduce energy consumption during the combined maximum hours of the region, grid capacity hours, as well as in the implementation of electricity demand response events. Universal forecasting methods (black-box methods) based on linear regression, baseline, neural network, autoregressive, triple exponential smoothing (Holt-Winters model), autoregressive model with seasonality support (SARIMA) and methods based on ensemble models are considered as examples. A scheme for integrating computational and analytical models with IoT platforms InfluxData and EMS INSYTE is proposed. The new architecture provides the execution of various short-term energy forecasting models on the analyst server. The results of experimental research of the performance of custom analytics in the composition of IoT platforms in the implementation of ventilation predictive control scenarios for the day ahead are presented.